Coming to Enemy Scout: the sixty-second card

Between the map pick and the knife round you get about a minute. That is the whole window a pug player has to look at anything about the other team, and it has shaped everything we are about to ship.

This post is a preview of the next Enemy Scout update, what it changes, and why. It is rolling out to staff testers now and to the extension and web panel over the coming weeks.

The problem with scouting dashboards

Most scouting products, ours included until now, hand you a page of numbers: site percentages, pistol win rates, buy habits, per-player stats. It looks thorough. It is also unusable in sixty seconds, and there is a quieter problem underneath: a lot of those numbers describe the team your opponent happened to be on in their last game, not the player who is about to be in your lobby. In a pug the roster changes every match. The reads don't carry.

So we went back to the demos. Over the last two weeks we took hundreds of parsed FACEIT demos and thousands of rounds and asked one question of every stat we could compute: if we measure this about a player in one match, does it still hold in their next one? We also took every Enemy Scout read we had made before a real match and graded it against the demo of the match that was actually played.

We are not going to publish the recipe. But the headline is worth saying out loud: most team-level tendencies do not survive contact with the next pug, and a small set of per-player habits do. That set is what the new card is built from.

What the card looks like

Two cards, one for each side you might start on. Each one is a single radar with the five enemy players on it, and five short rows underneath.

• Who has the AWP. The one thing that changes how you peek every angle facing his lane.
• Who takes the first fight and who throws the utility. The player whose smokes landing means the execute is on.
• Where each of them is early in the round, by name. Not a heatmap blob. "P2 holds window" is something you can pre-aim; a warm smudge on a map is not.
• One line at the top telling you the single most useful thing to do this match, and one grey line under it telling you what to do if that read turns out wrong.

There are no percentages on the card. Every claim carries the number of demos behind it instead, and every marker is drawn solid or dashed depending on how consistent the player has been. We tested those markers against matches played after the reads were made, and the solid ones earn their confidence.

No AI required

The card is computed straight from the demos, with no AI in the loop. That matters for two reasons. It is instant and it costs nothing to generate, so it can go to more players than the AI reports could. And it is deterministic: the same demos always produce the same card, so it can be graded, and it can be wrong in ways we can find and fix.

The AI Team Read is not going away. It becomes a button. When you want the longer analyst-style briefing, you ask for it; it no longer runs on its own every time a lobby loads.

What already shipped underneath

Some of this work is already live, because the card is only as good as the demos it reads.

• The demo pipeline was re-verified end to end. We found and fixed a round-counting defect and re-parsed every demo on file.
• Bomb-site labels now come straight from the game's own callouts instead of from map coordinates. Position reads on every map are more precise as a result, and a couple of maps that were being mislabeled are now right.
• Every position read now carries the actual callout name, so the card can say "Connector" rather than "somewhere around mid".

If you have been using the Team Read, the existing panel stays, with a cleaner layout: the verdicts up front, the detail folded underneath.

When

Staff testers have the card now. Pro and Coach get it next, in the extension side panel and the web panel, with the same token path free users already use for scouts. We will post again when it is live, along with the first accuracy numbers from real lobbies.

If you want to be in the first wave, the extension is where it lands first.